Skip to main content

Johannes Ballé: Density Modeling of Images using a Generalized Divisive Normalization Transformation

Movies Preview

texts
Johannes Ballé: Density Modeling of Images using a Generalized Divisive Normalization Transformation


Published February 19, 2016
SHOW MORE


Talk by Johannes Ballé, who is visiting from Eero Simoncelli’s lab at NYU; given to the Redwood Center for Theoretical Neuroscience at UC Berkeley.

Abstract
We introduce a parametric nonlinear transformation for jointly Gaussianizing patches of natural images. The transformation is differentiable, can be efficiently inverted, and thus induces a density model. It generalizes and performs better than several previous image models such as ICA, radial Gaussianization, and ISA. Model samples are visually similar to natural image patches. We use the model for image restoration, and show that it can be cascaded to build nonlinear hierarchies analogous to multiscale representations.


Language English
Collection opensource_media

comment
Reviews

There are no reviews yet. Be the first one to write a review.
SIMILAR ITEMS (based on metadata)
Community Media
texts
eye 97
favorite 0
comment 0
Community Texts
texts
eye 22
favorite 0
comment 0
Community Media
texts
eye 15
favorite 0
comment 0
Community Media
by vincent charlebois
data
eye 46
favorite 0
comment 0
Community Texts
by Woodrow Wilson Bledsoe
texts
eye 28
favorite 0
comment 0
Arxiv.org
by Stephane Coen; Hamish G. Randle; Thibaut Sylvestre; Miro Erkintalo
texts
eye 36
favorite 0
comment 0
Source: http://arxiv.org/abs/1211.1697v2
Community Media
texts
eye 21
favorite 0
comment 0
Arxiv.org
texts
eye 20
favorite 0
comment 0
Source: http://arxiv.org/abs/0810.3304v1